diff --git a/README.md b/README.md index 7e78d9b..03a4fe4 100644 --- a/README.md +++ b/README.md @@ -400,7 +400,10 @@ If you like what I share, please support me with [PayPal](https://paypal.me/jake 1. Add SAM3D Mesh Sequence From Video node to get mesh sequence from video using SAM3D. Need to install [ComfyUI-SAM3DBody](https://github.com/PozzettiAndrea/ComfyUI-SAM3DBody). - 2026-02-01 - v2.5.4 -1. Update SAM3D Mesh Sequence From Video node: Align the mesh sequence to the video instead of aligning along the bbox center. +1. Update SAM3D Mesh Sequence From Video node: Align the mesh sequence to the video instead of aligning along the creation center. + +- 2026-02-02 - v2.5.5 +1. Update SAM3D Mesh Sequence From Video node: set z offset to 0 for mesh sequence. ## Installation 1. `git clone https://github.com/jakechai/ComfyUI-JakeUpgrade` into the `custom_nodes` folder diff --git a/nodes/jake_node_3d_sam3dseq.py b/nodes/jake_node_3d_sam3dseq.py index d8466e5..665044a 100644 --- a/nodes/jake_node_3d_sam3dseq.py +++ b/nodes/jake_node_3d_sam3dseq.py @@ -436,23 +436,38 @@ def align_models_to_fixed_camera(frames_data, faces, reference_frame=0): aligned_vertices[i] = ref_vertices.copy() continue - # 应用对齐变换：V_i' = V_i + C_i - C_ref - # 注意：这里假设焦距相同，如果需要处理变焦，需要额外的缩放 + # 应用对齐变换 + # X和Y方向：V_i' = V_i + C_i - C_ref + # Z方向：V_i' = V_i + if focal_length is not None and ref_focal_length is not None and focal_length != ref_focal_length: - # 如果需要处理焦距变化 + # 如果需要处理焦距变化，先计算缩放因子 scale = ref_focal_length / focal_length - aligned_vertices[i] = (vertices + cam_t) * scale - ref_cam_t + + # 对X和Y方向应用缩放 + aligned_vertices[i, :, 0] = (vertices[:, 0] + cam_t[0] - ref_cam_t[0]) * scale + aligned_vertices[i, :, 1] = (vertices[:, 1] + cam_t[1] - ref_cam_t[1]) * scale + + # Z方向不缩放 + aligned_vertices[i, :, 2] = vertices[:, 2] + if i < 3: # 打印前几帧的调试信息 - print(f"[Alignment] Frame {i}: scaled by {scale:.3f} (focal: {focal_length:.1f} -> {ref_focal_length:.1f})") + print(f"[Alignment] Frame {i}: XY scaled by {scale:.3f} (focal: {focal_length:.1f} -> {ref_focal_length:.1f})") else: - # 简单的相机位置对齐 - aligned_vertices[i] = vertices + cam_t - ref_cam_t + # X和Y方向：V_i' = V_i + C_i - C_ref + aligned_vertices[i, :, 0] = vertices[:, 0] + cam_t[0] - ref_cam_t[0] + aligned_vertices[i, :, 1] = vertices[:, 1] + cam_t[1] - ref_cam_t[1] + + # Z方向：V_i' = V_i + aligned_vertices[i, :, 2] = vertices[:, 2] valid_frames.append(i) # 调试信息 if i < 3: # 只打印前3帧的详细信息 - print(f"[Alignment] Frame {i}: cam_t={cam_t.flatten()}, aligned offset={-cam_t.flatten() + ref_cam_t.flatten()}") + print(f"[Alignment] Frame {i}: cam_t={cam_t.flatten()}") + print(f" XY offset: +cam_t - ref_cam_t = {cam_t[0]-ref_cam_t[0]:.3f}, {cam_t[1]-ref_cam_t[1]:.3f}") + print(f" Z offset: 0.000") print(f"[Alignment] Completed: {len(valid_frames)} valid frames aligned to reference camera") @@ -555,7 +570,7 @@ class SAM3DMeshSequenceFromVideo_JK: OUTPUT_NODE = True def __init__(self): - """初始化输出目录，参考jake_node_3d_viewer.py""" + """初始化输出目录""" # 检查依赖是否可用 available, error_msg = check_sam3d_dependencies() if not available: @@ -574,7 +589,7 @@ class SAM3DMeshSequenceFromVideo_JK: return Path(self.output_dir) / self.tmp_output_dir_name def clean_file_path(self, file_path: str) -> str: - """清理文件路径，去除可能的引号（参考jake_node_3d_viewer.py）""" + """清理文件路径，去除可能的引号""" if not file_path: return "" @@ -811,8 +826,8 @@ class SAM3DMeshSequenceFromVideo_JK: processed_frames += 1 - # 每处理10帧显示一次进度 - if (frame_idx + 1) % 10 == 0 or (frame_idx + 1) == num_frames: + # 每处理30帧显示一次进度 + if (frame_idx + 1) % 30 == 0 or (frame_idx + 1) == num_frames: elapsed = time.time() - start_time_total fps_rate = processed_frames / elapsed if elapsed > 0 else 0 print(f"[VideoFramesToMesh] Progress: {frame_idx + 1}/{num_frames} frames, " @@ -822,7 +837,7 @@ class SAM3DMeshSequenceFromVideo_JK: if processed_frames == 0: return (f"错误: 没有成功处理任何帧",) - # 6. 根据参数决定是否应用模型对齐 + # 6. 默认应用模型对齐 if align_to_reference_camera: print(f"[VideoFramesToMesh] Aligning models to fixed camera (reference frame: {reference_frame})...") try: @@ -851,7 +866,7 @@ class SAM3DMeshSequenceFromVideo_JK: for d in frames_data], axis=0) valid_frames = [i for i, d in enumerate(frames_data) if d["success"]] - # 7. 总是应用时间平滑（如果有多于1帧） + # 7. 默认应用时间平滑（如果有多于1帧） if len(valid_frames) > 1: print(f"[VideoFramesToMesh] Applying temporal smoothing (method: {smoothing_method}, sigma: {smoothing_sigma})") @@ -866,14 +881,14 @@ class SAM3DMeshSequenceFromVideo_JK: print(f"[VideoFramesToMesh] Temporal smoothing completed") - # 8. 应用坐标变换 + # 8. 默认应用坐标变换 if coordinate_transform != "none": print(f"[CoordinateTransform] Applying coordinate transformation: {coordinate_transform}") # 变换顶点序列 aligned_vertices = apply_coordinate_transform(aligned_vertices, coordinate_transform) - # 9. 总是保存为 SMPL 兼容格式 + # 9. 默认保存为 SMPL 兼容格式 MeshSequenceBinaryFormat.save_smpl_compatible( vertices_sequence=aligned_vertices, faces=faces, @@ -887,7 +902,6 @@ class SAM3DMeshSequenceFromVideo_JK: print(f"[VideoFramesToMesh] Model alignment: enabled (reference frame: {ref_cam_params.get('reference_frame', 0)})") print(f"[VideoFramesToMesh] Coordinate transformation: {coordinate_transform}") print(f"[VideoFramesToMesh] Temporal smoothing: enabled ({smoothing_method}, sigma={smoothing_sigma})") - print(f"[VideoFramesToMesh] Output file: {output_path}") # 只返回完整路径 return (str(output_path),) diff --git a/pyproject.toml b/pyproject.toml index 8cd5e6d..dfc1d40 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -1,7 +1,7 @@ [project] name = "comfyui-jakeupgrade" description = "ComfyUI workflow customization by Jake." -version = "2.5.4" +version = "2.5.5" license = {file = "LICENSE"} dependencies = ["opencv-python", "piexif", "torch", "torchvision", "numpy", "pillow", "simpleeval", "PyYAML", "toml"]